The single most valuable minute you will ever have with a customer is the minute they are paying you. They have already decided to trust you, already reached for a card or handed cash to a courier, already crossed the psychological line from considering to buying. Every dollar of acquisition cost you were ever going to spend on them is already spent. Intent will never be higher and access will never be cheaper. Yet most businesses let that minute pass with a single transaction, when it is the one moment engineered to hold several.
Alex Hormozi's clearest contribution in $100M Money Models is naming the sequence that spends this minute properly. Land the first yes, then present an upsell at peak intent immediately after — not next week in an email, but right there, while the commitment is warm. Offer a downsell, a smaller yes, to the people who decline the upsell, so a "no" to the big thing does not become a "no" to everything. And layer continuity, recurring revenue, underneath, so the relationship compounds instead of resetting to zero after each sale. The frame is simple and correct: no lead should reach the point of purchase and cross it without transacting at some level.
That is a genuinely useful discipline. It is also the exact place where the guru version turns predatory, because the same mechanics that let you serve a customer better let you strip-mine a moment of commitment. So the argument I want to make is narrow: the point-of-sale sequence is real money and you should build it deliberately — but every offer in it has to answer a need the purchase itself created, or it extracts once and poisons the well you were going to draw from for years.
Why the moment of purchase is near-free margin
Start with the economics, because they are more extreme than the tactics suggest.
Say you spend 40 to acquire a customer and your first product sells for 60 at a 50% gross margin. You made 30 in gross profit and you are 10 underwater after acquisition cost. On that math you need a second purchase, someday, just to break even — and "someday" means another round of attention, another reason to come back, more cost to re-engage.
Now look at an upsell presented at the moment of that first purchase. The acquisition cost is already sunk. You are not paying to reach this person; they are standing at your counter. The marginal cost of presenting the next offer is essentially zero, and the marginal cost of delivering it is only whatever the second product actually costs to fulfill. If 25% of buyers take an upsell that carries 50 of margin, that is an extra 12.50 of gross profit per original customer — enough, in this example, to flip the whole cohort from 10 underwater to 2.50 positive on day one. Nothing about the product changed. You simply stopped forgoing margin that was sitting in front of you.
This is the real reason "get paid before you spend on the next acquisition" works as a growth lever. It is not a mindset. It is that cash velocity — how fast a customer returns the cash you spent to acquire them — is mathematically independent of your margin and your market size. Two businesses with identical unit economics grow at wildly different speeds if one recoups its CAC at the register and the other recoups it over six months. The first funds the next acquisition out of the last one's proceeds. The second funds growth out of a bank account, which runs dry.
The downsell captures a different quantity: the option value of a no. When someone declines your upsell, they have not usually rejected you — they have rejected a price or a size of commitment at that instant. A downsell converts some of those refusals into a smaller yes, which is near-pure margin on a customer you already have, and, more importantly, keeps them inside the relationship where they can ascend later. A no that walks out the door is worth nothing. A no that becomes a 19 yes is worth 19 plus the entire future you have kept alive.
The same minute can be spent or strip-mined
Here is where I part from the tactical version. The properties that make the moment of purchase valuable — commitment, momentum, lowered guard — are exactly the properties that make it exploitable. A person mid-transaction will say yes to things they would reject with a clear head. You can use that to serve them faster or to fleece them faster, and the mechanics look identical on the day. They diverge completely on the second purchase.
The distinction that matters is whether the next offer answers a need the purchase itself created.
When someone buys a training program, the act of buying it creates a new, real problem: now they have to implement it, and implementation is where most of them will fail. An upsell of coaching, done-with-you support, or a tool that removes the hardest step is not an add-on. It is the answer to the problem the sale just handed them. Taking it makes them more likely to reach the dream outcome they bought the first thing for. That upsell raises the perceived value of the original purchase in hindsight, and it deepens trust, because it shows you understand where they are about to get stuck.
Contrast the order bump that has nothing to do with the outcome — the pre-checked box, the "customers also protect their purchase for just 7.99," the irrelevant warranty on a digital product. That converts too. It converts because the guard is down, not because it serves a need. It extracts once. And it teaches the customer something precise: that the moment they trusted you was a moment you were waiting to exploit. You will feel that lesson later, as a refund, a chargeback, a subscription cancelled the instant it is noticed, a review that uses the word "sneaky." You did not add value. You added a tax, and taxes get evaded and resented.
This is not a soft ethical preference. It is the arithmetic of the money model catching up with you. The entire reason the sequence beats a single sale is that it assumes a relationship — a customer who returns, ascends, and refers. An extractive upsell trades that relationship for one incremental margin capture. You are selling an annuity to buy a coupon.
Continuity is only revenue you can hold
The layer where this goes most wrong is continuity, because a subscription launders a one-time extraction into a recurring number that looks like a business.
Booked recurring revenue is not revenue until it is retained. I have argued this at length in Recurring Revenue Is a Trap Without the Retention Math, and it is the exact failure mode a money model invites: you get so good at converting the moment of purchase into a subscription that you stop asking whether the subscription survives the second billing cycle. If it does not, continuity is not compounding. It is a bucket you refill every month while it leaks, and the leak stays invisible on the top-line number until acquisition slows and the base drains.
Do the arithmetic before you celebrate. If you churn 10% of subscribers a month, your average customer stays ten months, so a 99-a-month plan is worth about 990 in lifetime revenue, not the 1,188 that annualizing day-one signups implies — and if churn is really 15%, the average lifetime is under seven months and the figure drops to roughly 660. Price your acquisition against the number you can hold, not the number a fresh subscriber flatters you with. Continuity earned by genuinely solving a recurring problem retains. Continuity manufactured by making cancellation hard, or by auto-enrolling people who did not understand they subscribed, churns the moment attention returns to it — and it churns angry.
Whose need are you serving in that minute?
Behind all of this is a design choice about who the moment of purchase is for. When you build the upsell, you are deciding whether the next thing the customer sees serves their next need or your next quarter. Those are not always opposed, but when they are, the direction you resolve them in becomes the character of the whole business — which is the argument in Whoever Pays Decides What You Build. Optimize the upsell purely for take-rate and you will end up building for the version of the customer who is easiest to extract from in a weak moment. That customer does not come back.
I can make this concrete from Kommerce. We run commerce infrastructure for cash-on-delivery markets, where a customer places an order and pays the courier at the door — which means the merchant has extended unusual trust before a cent changes hands, and so has the buyer. At the moment a merchant onboards and processes their first orders, a real problem is created immediately: a meaningful share of COD orders fail — the customer isn't home, refuses at the door, or the address is wrong — and every failure costs the merchant shipping both directions and locks up cash. The natural, honest upsell there is COD-risk tooling: delivery-attempt optimization, buyer-scoring that flags likely refusals, premium logistics that raise the success rate. That offer exists because the first purchase surfaced the exact pain it solves. It makes the merchant more successful at the thing they already bought us for, which makes them retain — the only continuity that counts.
The extractive version is available too, and in a low-trust market it is both more tempting and more damaging. A checkout add-on the merchant doesn't need, a "priority support" tier that just restores support we quietly degraded, an auto-renewing fee buried in onboarding. Every one of those would convert in the moment. Every one would surface later as a merchant who felt handled — and in COD, a merchant who feels handled tells the other merchants, because these are small, connected markets where trust is the scarce asset and reputation travels fast. The same trust that makes cash-on-delivery function at all is the trust an extractive upsell spends. You cannot afford to spend it.
The move
Build the sequence, deliberately, and hold each offer to a single test.
- Design one upsell, the single most relevant one. Ask what problem the purchase creates — the thing the customer will hit the moment after your product works — and offer precisely the thing that removes it. One offer, presented at the register, matched to a created need. Not a menu; the one that makes the original purchase more likely to succeed.
- Design a graceful downsell for the no. A smaller yes matched to a real, smaller need or a lower commitment level — enough to keep the relationship alive and capture the margin you already paid to earn. Never a nagged, cornered, or disguised yes.
- Count only continuity you can retain. Compute your monthly churn, derive the real average lifetime, and price acquisition against lifetime value you can hold — not the annualized figure a day-one subscriber suggests.
The point of sale is the one moment where serving the customer more and earning more point in the same direction — if the next offer answers a need the sale just created. Spend that minute. Do not strip-mine it.